Availableonlineatwww.sciencedirect.com Science Direct E噩≈RS ELSEVIER Joumal of the European Ceramic Society 28(2008)2801-2807 www.elsevier.comlocate/jeurceramsoc Electrophoretic deposition in the production of Sic/Sic composites for fusion reactor applications Sasa Novak a, * Katja Rade a, Katja Konig a, Aldo R. Boccaccinib Department for Nanostructured Materials, Jozef Stefan Institute, Ljubljana, Slovenia b Department of Materials, Imperial College London, South Kensington Campus, London SW72BP UK Received 5 January 2008: received in revised form 25 March 2008: accepted 4 April 2008 Available online 2 June 2008 Abstract This paper presents the results of a study aimed at developing a technique for the infiltration of a SiC-fibre fabric with a low-activation SiC-based matrix material by aid of electrophoretic deposition from aqueo ensions. To achieve the best possible particle packing in the infiltrated matrix and hence to minimise the shrinkage during drying, the effect of the suspensions composition was analysed. Besides the ph change, different types of surfactants-PEL, CTAB, citric acid and Dolapix-were included in the investigation. As a result, during the deposition onto a metallic electrode, the best deposits were obtained with the addition of CTAb or by increasing the ph, while the infiltration of the Sic-fibre fabric wa more effective with negatively charged particles in the suspension o 2008 Elsevier Ltd. all rights reserved. Keywords: Silicon carbide; Aqueous suspensions; Surfactants; Electrophoretic deposition; Electrophoretic infiltration; SiC/SiC composites; Fusion 1. ntroduction limit the potential for SiC-based composites to be seriously con- sidered as suitable materials for the next generation of fusion Continuous Sic-fibre-reinforced SiC-matrix composites reactors. For this reason, efforts are being made worldwide (SiC/SiC) are recognised as promising materials to solve the demanding issues related to processing of SiC/Sic demanding applications due to their ability to resist composites. One key issue is related to the feasibility of achiev- conditions, for example, in heat-engine components in full infiltration of the sic-fibre fabric with a low-activation propulsion and in the structural parts of future fusion reactors. 2 matrix material The aim of this investigation is to develop a Sic-based compos- The infiltration of Sic-fibre woven fabric with a sic-matrix ite that will effectively substitute the currently favoured ferrous material has been undertaken by various techniques, notably materials, which are proposed to be used in future fusion reac- chemical vapour infiltration(CVI) and polymer infiltration and tors mainly due to the high degree of present technological pyrolysis(PIP). Unfortunately, these are very slow and costly development of relevant alloys. On the contrary, the fusion- processes and/or result in an incomplete filling of the gaps relevant SiC-based composites are the least well developed between the fibres in the tows. In addition to CVI and PIP, among the candidate materials for a reactor; however, using which result in incomplete filling and the formation of highly SiC/SiC composites for the blanket structural component could amorphous SiC, further attempts have been based on using lead to a significant increase in the maximum operating tem- ceramic routes. Among them, the recently introduced NT perature and, moreover, the material would not decay under process, based on a transient eutectic-phase route, , seems neutron irradiation to produce long-lived radioactive waste. to be the best suited to meet the requirements for a mate- There are. however some critical issues. such as insufficient rial to be used in a fusion reactor The infiltration in this hermeticity, swelling and various technological obstacles, that process is performed by slip infiltration, followed by assisted liquid-phase sintering of the SiC matrix using Al2O3 and Y2O3 as the sintering aids. Most recently, there have orresponding author. been attempts to avoid or at least minimise these additives E-lmail address: sasa. novak @ijssi(S. Novak) with the aim to produce a material with the lowest pos 0955-2219/S-see front matter o 2008 Elsevier Ltd. All rights reserved. doi: 10. 1016/j-jeurceramsoc. 2008.04.004
Available online at www.sciencedirect.com Journal of the European Ceramic Society 28 (2008) 2801–2807 Electrophoretic deposition in the production of SiC/SiC composites for fusion reactor applications Sasa Novak ˇ a,∗, Katja Rade a, Katja Konig ¨ a, Aldo R. Boccaccini b a Department for Nanostructured Materials, Joˇzef Stefan Institute, Ljubljana, Slovenia b Department of Materials, Imperial College London, South Kensington Campus, London SW7 2BP, UK Received 5 January 2008; received in revised form 25 March 2008; accepted 4 April 2008 Available online 2 June 2008 Abstract This paper presents the results of a study aimed at developing a technique for the infiltration of a SiC-fibre fabric with a low-activation SiC-based matrix material by aid of electrophoretic deposition from aqueous suspensions. To achieve the best possible particle packing in the infiltrated matrix and hence to minimise the shrinkage during drying, the effect of the suspension’s composition was analysed. Besides the pH change, different types of surfactants – PEI, CTAB, citric acid and Dolapix – were included in the investigation. As a result, during the deposition onto a metallic electrode, the best deposits were obtained with the addition of CTAB or by increasing the pH, while the infiltration of the SiC-fibre fabric was more effective with negatively charged particles in the suspension. © 2008 Elsevier Ltd. All rights reserved. Keywords: Silicon carbide; Aqueous suspensions; Surfactants; Electrophoretic deposition; Electrophoretic infiltration; SiC/SiC composites; Fusion 1. Introduction Continuous SiC-fibre-reinforced SiC-matrix composites (SiC/SiC) are recognised as promising materials for many demanding applications due to their ability to resist extreme conditions, for example, in heat-engine components in aerospace propulsion and in the structural parts of future fusion reactors.1,2 The aim of this investigation is to develop a SiC-based composite that will effectively substitute the currently favoured ferrous materials, which are proposed to be used in future fusion reactors mainly due to the high degree of present technological development of relevant alloys. On the contrary, the fusionrelevant SiC-based composites are the least well developed among the candidate materials for a reactor; however, using SiC/SiC composites for the blanket structural component could lead to a significant increase in the maximum operating temperature and, moreover, the material would not decay under neutron irradiation to produce long-lived radioactive waste.3 There are, however, some critical issues, such as insufficient hermeticity, swelling and various technological obstacles, that ∗ Corresponding author. E-mail address: sasa.novak@ijs.si (S. Novak). limit the potential for SiC-based composites to be seriously considered as suitable materials for the next generation of fusion reactors.4–6 For this reason, efforts are being made worldwide to solve the demanding issues related to processing of SiC/SiC composites. One key issue is related to the feasibility of achieving full infiltration of the SiC-fibre fabric with a low-activation matrix material. The infiltration of SiC-fibre woven fabric with a SiC-matrix material has been undertaken by various techniques, notably chemical vapour infiltration (CVI) and polymer infiltration and pyrolysis (PIP). Unfortunately, these are very slow and costly processes and/or result in an incomplete filling of the gaps between the fibres in the tows. In addition to CVI and PIP, which result in incomplete filling and the formation of highly amorphous SiC, further attempts have been based on using ceramic routes. Among them, the recently introduced NITE process, based on a transient eutectic-phase route,7,8 seems to be the best suited to meet the requirements for a material to be used in a fusion reactor. The infiltration in this process is performed by slip infiltration, followed by pressureassisted liquid-phase sintering of the SiC matrix using Al2O3 and Y2O3 as the sintering aids. Most recently, there have been attempts to avoid or at least minimise these additives with the aim to produce a material with the lowest pos- 0955-2219/$ – see front matter © 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jeurceramsoc.2008.04.004
2802 S Novak et al. /Journal of the European Ceramic Society 28(2008)2801-2807 ible neutron activation, and to develop dense, high-purity using a ZetaProbe analyser( Colloidal Dynamics, USA), while a ZetaPals instrument(Brookhaven, USA)was employed to According to Nannetti et al., hybrid techniques offer new analyse the Zp of the fibres. Surface charge was modified possibilities for the production of high-purity gas-impermeable with hydrochloric acid, citric acid (CA, Johnson Matthey SiC/SiC composites By filling approximately one third of the GmbH, Germany), tetramethyl-ammoniumhydroxide (Tmah) inter-bundle voids with SiC powder before PIP, they achieved polyethylene-imine (PEl 10.000, Alfa Aesar, Germany ), In our investigation we have looked at the feasibility of elec- Dolapix CE64(Zschimmer Schwarz, Germany Aldrich)or an cetyltrimethylammonium bromide(CTAB, Sigma rophoretic deposition(EPD)as a potential technique to fully The EPD experiments were performed with suspensions con- infiltrate SiC-fibre fabric with Sic powder. The aim of the taining 25 wt % o of solids at a constant dc voltage of 60V study was to achieve the highest possible packing density of the for 5 min using steel or copper electrodes. The electrodes owder-infiltrated preform that will be subsequently densified (20 mm x 20 mm x 0.5 mm)were positioned vertically in the either by Plp or another appropriate technique. EPD cell at a distance of 2 cm. In the experiments, the Sic-fib EPD is a commonly used process for producing coatings and woven fabric was used as the deposition electrode or placed self-standing ceramic parts; it is generally recognised as a fast, front of a steel electrode. The porosity of the deposits due to bub- simple and efficient ceramic processing technique. 0-14 Several ble formation was minimised by placing a cellulose membrane in successful attempts have been reported on the use of EPD for the front of the steel cathode or by using copper as the anode. Before production of ceramic fibre-matrix composites. It has been deposition the Sic-fabrics were rinsed with acetone and dis- shown that, depending on the electrical conductivity of the fibres, tilled water or pre-treated with a sodium dioctyl-sulfosuccinate the fibre-preform can either be used directly as the deposition (SDOSS)solution. 20 The progress of the deposition was mon- electrode or it can be positioned in front of the electrode. itored by the change in the current. The bulk deposits were In spite of the clear potential and effectiveness of the EPd evaluated in terms of the final weight and the solids content, process, the mechanisms and kinetics of the formation of the while the degree of particle packing was assessed with a scan- deposit are still not fully understood. Most papers report on ning electron microscope. The microstructures of the infiltrated results achieved by a trial-and-error approach, which frequently samples were observed in the green state using optical and elec leads to uncertain reproducibility. However, in our study we used tron microscopy the approach of a thorough analysis of the effect of the aque ous suspensions' composition and their electrokinetic behaviour 3. Results and discussion within a wide pH and zeta-potential range. This experimental approach should enable the reproducible deposition of a coat- 3. 1. Characteristics of the suspensions ing or formation of a matrix material with high green density as well as the infiltration of various fibre fabrics The use of a The properties of the aqueous SiC suspensions for elec- water-based system rather than the previously studied ethanol- trophoretic deposition were analysed in undiluted form, i.e based systems', was expected to result in some advantages containing 25 wt. solids. As shown in Fig. 1, the natural lue to the higher dielectric constant of water and the wider range pH(pH4)of the SiC suspension nearly matches the IEP, of suitable surfactants. Moreover, aqueous suspensions are also which explains why the suspension is very unst more attractive because unlike ethanol-based suspensions they with HCl resulted in a significant increase in the conductivity are not sensitive to humid environments, which should result in and only slight increase in the ZP up to a maximum value of better reproducibility. The main objective of this investigation was therefore to tailor 11 mV at pH 2.8, where the stability of the suspension was still SiC suspension's properties with positive and negative particle net-surface charges in order to be able to adapt the suspensions properties for the infiltration of a particular type of Sic-fibre HCI TMAH fabric. Our work was primarily focused on a comprehensive study of the EPD process; the investigation of the subsequ uent densification of the SiC matrix produced was not part of this R 2. Experimental The substrate material used in this investigation was Tyrano 。8 SA SiC-fibre fabric(Ube Industries, Ltd, Japan). For the infil tration we used B-Sic powder BF12(H. Starck, Germany) 1234567 with an average grain size of 0.5 um. Aqueous suspensions ontaining 25 wt %o of powder were prepared by homogenise- Fig. 1. The influence of pH change on the zeta-potential (ZP) and the conduc. tion in a multidirectional mixer for 30 min. The zeta-potential tivity of an aqueous suspension of Sic powder(solids content: 25 wt%;pH ZP) of the powders was measured in undiluted suspensions adjusted by HCl and TMAH
2802 S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 sible neutron activation, and to develop dense, high-purity SiC. According to Nannetti et al.,9 hybrid techniques offer new possibilities for the production of high-purity gas-impermeable SiC/SiC composites. By filling approximately one third of the inter-bundle voids with SiC powder before PIP, they achieved less than 10% porosity; however, this is still unacceptably high. In our investigation we have looked at the feasibility of electrophoretic deposition (EPD) as a potential technique to fully infiltrate SiC-fibre fabric with SiC powder. The aim of the study was to achieve the highest possible packing density of the powder-infiltrated preform that will be subsequently densified either by PIP or another appropriate technique. EPD is a commonly used process for producing coatings and self-standing ceramic parts; it is generally recognised as a fast, simple and efficient ceramic processing technique.10–14 Several successful attempts have been reported on the use of EPD for the production of ceramic fibre-matrix composites.14–16 It has been shown that, depending on the electrical conductivity of the fibres, the fibre-preform can either be used directly as the deposition electrode or it can be positioned in front of the electrode. In spite of the clear potential and effectiveness of the EPD process, the mechanisms and kinetics of the formation of the deposit are still not fully understood. Most papers report on results achieved by a trial-and-error approach, which frequently leads to uncertain reproducibility. However, in our study we used the approach of a thorough analysis of the effect of the aqueous suspensions’ composition and their electrokinetic behaviour within a wide pH and zeta-potential range. This experimental approach should enable the reproducible deposition of a coating or formation of a matrix material with high green density as well as the infiltration of various fibre fabrics. The use of a water-based system rather than the previously studied ethanolbased systems17,18 was expected to result in some advantages due to the higher dielectric constant of water and the wider range of suitable surfactants. Moreover, aqueous suspensions are also more attractive because unlike ethanol-based suspensions they are not sensitive to humid environments, which should result in better reproducibility.19 The main objective of this investigation was therefore to tailor SiC suspension’s properties with positive and negative particle net-surface charges in order to be able to adapt the suspension’s properties for the infiltration of a particular type of SiC-fibre fabric. Our work was primarily focused on a comprehensive study of the EPD process; the investigation of the subsequent densification of the SiC matrix produced was not part of this work. 2. Experimental The substrate material used in this investigation was Tyrano SA SiC-fibre fabric (Ube Industries, Ltd., Japan). For the infiltration we used -SiC powder BF12 (H. Starck, Germany) with an average grain size of 0.5 m. Aqueous suspensions containing 25 wt.% of powder were prepared by homogenisation in a multidirectional mixer for 30 min. The zeta-potential (ZP) of the powders was measured in undiluted suspensions using a ZetaProbe analyser (Colloidal Dynamics, USA), while a ZetaPals instrument (Brookhaven, USA) was employed to analyse the ZP of the fibres. Surface charge was modified with hydrochloric acid, citric acid (CA, Johnson Matthey GmbH, Germany), tetramethyl-ammoniumhydroxide (TMAH), polyethylene-imine (PEI 10.000, Alfa Aesar, Germany), cetyltrimethylammonium bromide (CTAB, Sigma–Aldrich) or Dolapix CE64 (Zschimmer & Schwarz, Germany). The EPD experiments were performed with suspensions containing 25 wt.% of solids at a constant dc voltage of 60 V for 5 min using steel or copper electrodes. The electrodes (20 mm × 20 mm × 0.5 mm) were positioned vertically in the EPD cell at a distance of 2 cm. In the experiments, the SiC-fibre woven fabric was used as the deposition electrode or placed in front of a steel electrode. The porosity of the deposits due to bubble formation was minimised by placing a cellulose membrane in front of the steel cathode or by using copper as the anode. Before deposition the SiC-fabrics were rinsed with acetone and distilled water or pre-treated with a sodium dioctyl-sulfosuccinate (SDOSS) solution.20 The progress of the deposition was monitored by the change in the current. The bulk deposits were evaluated in terms of the final weight and the solids content, while the degree of particle packing was assessed with a scanning electron microscope. The microstructures of the infiltrated samples were observed in the green state using optical and electron microscopy. 3. Results and discussion 3.1. Characteristics of the suspensions The properties of the aqueous SiC suspensions for electrophoretic deposition were analysed in undiluted form, i.e., containing 25 wt.% solids. As shown in Fig. 1, the natural pH (pH ∼ 4) of the SiC suspension nearly matches the IEP, which explains why the suspension is very unstable. Titration with HCl resulted in a significant increase in the conductivity and only slight increase in the ZP up to a maximum value of 11 mV at pH 2.8, where the stability of the suspension was still Fig. 1. The influence of pH change on the zeta-potential (ZP) and the conductivity of an aqueous suspension of SiC powder (solids content: 25 wt.%; pH adjusted by HCl and TMAH).
S Novak et al. Journal of the European Ceramic Society 28 (2008)2801-2807 2803 (a) Table I Compositions and properties of the characteristic suspensions used in the EPD experiments and corresponding properties of wet deposits 0000 Suspension Deposit H Conductivity ZP wt %e -CTAB R 2.80.78 △PE|+HCI 3 CTAB 3.90.24 O CA+PEI 5(PED+CA 0.75+2280.86 Dolapix+ NaoH -60 6(PED+HCI 542 00.1020.3040.5060.708091 2.60.7 surfactant(wt % 8 (CA)+PEI 0.5+0.33.30.3 15 CTAB 15 12 (DCE64)+NaOH 0.5 -4962 EE至号 Fig. 2a also shows that the addition of citric acid to SiC suspension has a minor effect on the ZP, but it sigi icantly increases its conductivity. However, the conductivity decreased considerably with further additions of pEl, while the ZP increased from 15 to 40mV. This observation agrees with previous reports,23 which showed that the surfactant-addition 00.10.2030405060.70809 sequence has a large effect on the stability of the suspensions. surfactant (wt % The addition of the anionic deflocculant Dolapix CE64 increased the absolute value of the ZP up to -24 mv, which Fig. 2. The ZP(a)and conductivity change(b) due to the addition of different was further increased to-49 mV by adjusting the pHto 9: how surfactants to SiC suspensions(solids content: 25 wt %). The arrows show the ever, the subsequent conductivity increase was relatively higl effect of an additional pH change for a particular surfactant addition. (see Fig. l a and b) The above results are summarised in Table l and indicate that very low. This implies that a ph decrease will probably not high, positive zP values can be obtained with the addition of be sufficient for EPD and hence, to increase the positive net- 0.4% CTAB, 0.8% of PEI at pH& or through combined addition surface charge, the addition of a cationic surfactant is require of 0.5% CA and 0.3%o Pel, while high, negative values can be In contrast, the pH increase with TMAH or NaOH resulted in achieved by a pH increase to 9 or with the addition of Dolapix a much more pronounced effect on the ZP, giving a value of CE64 at pH 9. However, so slightly composition with characteristic ZP values ton these results, 50mV at pH 9. However, a further increase in the ph only strong increase in the conductivity too. Based increased the conductivity, whereas the ZP began to decrease in afurther step, we performed EPD experiments using a selected mne ig 2a and b illustrates the effect of the different surfactants the ZP and the conductivity of the SiC suspension. It is clear 3. 2. EPD experiments that the addition of the cationic surfactant ctab increased the ZP up to 48 mV without a significant change in the conductivity, The EPD experiments were performed by using suspensions while the addition of PEl resulted in a smaller increase in the with 25 wt %o of solids at selected characteristic points in the ZP ZP, reaching a maximum value of 35 mV with 0.8% PEL. As VS pH diagrams. In order to evaluate the quality of the deposits reported by Zhang et al. 21, 22 due to the low dissociation of steel electrodes were used in the first trials. The suspension PEI in the alkaline region, PEI does not stabilise the alkaline compositions and properties, related to the final result of the to, according to their reports, a significant increase in the ZP 21.22 First, an EPD experiment was performed using a susper In our measurements the ZP reached its highest value(44 mv) sion without any additive. As expected, due to the absence of only after the subsequent addition of 2 wt. of citric acid (at charged particles in the Sic suspension at its natural pH of 4.1 pH 2.8), whereas the conductivity was already very high. a no deposit was observed after 5 min at an applied voltage up better result was obtained when the pH of the PEl-containing to 60 V. When the pH of the suspension was adjusted to 2.8, suspension was adjusted with HCl: at a pH of 8 the ZP reached where the ZP was 1l mv, a very loose deposit was formed on 52 mv(the effect of the pHchanges are indicated with the dashed the cathode, but it slid from the electrode during removal from arrows in Fig. 2a and b). the
S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 2803 Fig. 2. The ZP (a) and conductivity change (b) due to the addition of different surfactants to SiC suspensions (solids content: 25 wt.%). The arrows show the effect of an additional pH change for a particular surfactant addition. very low. This implies that a pH decrease will probably not be sufficient for EPD and hence, to increase the positive netsurface charge, the addition of a cationic surfactant is required. In contrast, the pH increase with TMAH or NaOH resulted in a much more pronounced effect on the ZP, giving a value of −50 mV at pH 9. However, a further increase in the pH only increased the conductivity, whereas the ZP began to decrease slightly. Fig. 2a and b illustrates the effect of the different surfactants on the ZP and the conductivity of the SiC suspension. It is clear that the addition of the cationic surfactant CTAB increased the ZP up to 48 mV without a significant change in the conductivity, while the addition of PEI resulted in a smaller increase in the ZP, reaching a maximum value of 35 mV with 0.8% PEI. As reported by Zhang et al.,21,22 due to the low dissociation of PEI in the alkaline region, PEI does not stabilise the alkaline suspension, while any further addition of citric acid should lead to, according to their reports, a significant increase in the ZP.21,22 In our measurements the ZP reached its highest value (44 mV) only after the subsequent addition of 2 wt.% of citric acid (at pH 2.8), whereas the conductivity was already very high. A better result was obtained when the pH of the PEI-containing suspension was adjusted with HCl: at a pH of 8 the ZP reached 52 mV (the effect of the pH changes are indicated with the dashed arrows in Fig. 2a and b). Table 1 Compositions and properties of the characteristic suspensions used in the EPD experiments and corresponding properties of wet deposits No. Composition Suspension Deposit Additive wt.% pH Conductivity ZP wt.% solids 1 0 4 0.08 −3 / 2 HCl 2.8 0.78 14 / 3 CTAB 0.4 3.9 0.24 48 67 4 PEI 0.8 8 0.12 35 / 5 (PEI) + CA 0.75 + 2 2.8 0.86 44 54 6 (PEI) + HCl 0.8 8 0.3 52 64 7 CA 0.5 2.6 0.7 15 / 8 (CA) + PEI 0.5 + 0.3 3.3 0.3 40 59 9 NaOH 9 0.15 −54 66 10 TMAH 9 0.15 −50 67 11 DCE64 0.5 6.3 0.5 −24 / 12 (DCE64) + NaOH 0.5 9 0.74 −49 62 Fig. 2a also shows that the addition of citric acid to the SiC suspension has a minor effect on the ZP, but it significantly increases its conductivity. However, the conductivity decreased considerably with further additions of PEI, while the ZP increased from 15 to 40 mV. This observation agrees with previous reports22,23 which showed that the surfactant-addition sequence has a large effect on the stability of the suspensions. The addition of the anionic deflocculant Dolapix CE64 increased the absolute value of the ZP up to −24 mV, which was further increased to −49 mV by adjusting the pH to 9; however, the subsequent conductivity increase was relatively high (see Fig. 1a and b). The above results are summarised in Table 1 and indicate that high, positive ZP values can be obtained with the addition of 0.4% CTAB, 0.8% of PEI at pH 8 or through combined addition of 0.5% CA and 0.3% PEI, while high, negative values can be achieved by a pH increase to 9 or with the addition of Dolapix CE64 at pH 9. However, some of these compositions show a strong increase in the conductivity too. Based on these results, in a further step, we performed EPD experiments using a selected composition with characteristic ZP values. 3.2. EPD experiments The EPD experiments were performed by using suspensions with 25 wt.% of solids at selected characteristic points in the ZP vs. pH diagrams. In order to evaluate the quality of the deposits, steel electrodes were used in the first trials. The suspensions’ compositions and properties, related to the final result of the EPD experiments, are listed in Table 1. First, an EPD experiment was performed using a suspension without any additive. As expected, due to the absence of charged particles in the SiC suspension at its natural pH of 4.1, no deposit was observed after 5 min at an applied voltage up to 60 V. When the pH of the suspension was adjusted to 2.8, where the ZP was 11 mV, a very loose deposit was formed on the cathode, but it slid from the electrode during removal from the suspension
S Novak et al. /Journal of the European Ceramic Society 28(2008)2801-2807 (a) (b) Fig. 3. SEM micrographs of deposits formed from a CTAB-stabilised suspension on a steel electrode(a and b), macroscopic view of the deposit formed on the copper electrode(c and d). As expected, a firm deposit formed on the cathode from the Dolapix is well dissociated and helps to produce dense and firm uspension with the addition of 0.5 wt. CTAB, characterised deposits by a high ZP. As illustrated in Fig 3a, the deposit contained large The above results suggest that for electrophoretic deposition deposition electrode, which led to a firm and non-porous deposit the deposits e apix and. t sions for cathodic deposition were channel-like pores due to water electrolysis, while the particles the most appropriate susp in the bulk were densely and homogeneously packed(Fig. 3b). In those containing CTAB, while for deposition on an anode, the further EPD experiments the presence of bubbles in the deposits addition of Do an increase in the pH seem to give the was prevented by placing a porous membrane in front of the best results, since they lead to a high particle-packing density in (Fig. 3c and d). By measuring the weight change during the in the deposit made by EPD increased to 67 wt% comparing to 3.3. Infiltration of Sic-fibre woven fabric the initial 25% solids in the starting suspension In the next set of experiments we explored the electrophoretic The EPD experiments using suspensions with the addition infiltration of SiC-fibre woven fabrics by SiC particles Based of PEI were less successful; no deposit was retained on the on the above results we selected the two compositions with the electrode after its removal from the suspension. In spite of the highest ZP: the suspension of positively charged particles with ZP increase, the further addition of 2 wt% of citric acid did a 0.5% CTAB addition(48 mV)and the suspension with neg- not significantly improve the deposition: the deposit was loose atively charged particles and Dolapix addition(-49mV).For and weak and, accordingly, the weight change during drying better understanding of the infiltration, the ZP of the Sic fibre revealed low solids content in the fresh deposit (54 wt %).A was also analysed, as shown in Fig 4. The as-received fibres used firm and dense deposit(64 wt % )was obtained from the suspen sion containing the 0.8%o PEI and the pH adjusted to 8 by HCl No deposit was obtained from the suspension containing citric id, which contrasts with the successful deposition in ethanol revIous In further experiments the deposit was formed on the anode 2-20 using negatively charged powder suspensions. In this case the presence of bubbles was prevented by using a Cu electrode R which consumes the oxygen formed during the electrolysis The alkaline suspension with the addition of TMA resul in a firm and dense deposit, containing 67 wt %o solids. A sim- non treated SDoss treated ilar result was obtained if the ph was adjusted to 9 by adding NaOH. In contrast, no deposit was formed from the suspen- Fig 4. Zeta-potential of non-treated (a)and (b)SDosS-treated Sit sion with Dolapix until the pH was adjusted to 9, where the function of pH(the fibres were cut and crushed for the analysed Sic fibres as a
2804 S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 Fig. 3. SEM micrographs of deposits formed from a CTAB-stabilised suspension on a steel electrode (a and b), macroscopic view of the deposit formed on the copper electrode (c and d). As expected, a firm deposit formed on the cathode from the suspension with the addition of 0.5 wt.% CTAB, characterised by a high ZP. As illustrated in Fig. 3a, the deposit contained large channel-like pores due to water electrolysis, while the particles in the bulk were densely and homogeneously packed (Fig. 3b). In further EPD experiments the presence of bubbles in the deposits was prevented by placing a porous membrane in front of the deposition electrode, which led to a firm and non-porous deposit (Fig. 3c and d). By measuring the weight change during the drying of the deposit, it was determined that the solids content in the deposit made by EPD increased to 67 wt.% comparing to the initial 25% solids in the starting suspension. The EPD experiments using suspensions with the addition of PEI were less successful; no deposit was retained on the electrode after its removal from the suspension. In spite of the ZP increase, the further addition of 2 wt.% of citric acid did not significantly improve the deposition: the deposit was loose and weak and, accordingly, the weight change during drying revealed low solids content in the fresh deposit (54 wt.%). A firm and dense deposit (64 wt.%) was obtained from the suspension containing the 0.8% PEI and the pH adjusted to 8 by HCl. No deposit was obtained from the suspension containing citric acid, which contrasts with the successful deposition in ethanol reported previously.17,18 In further experiments the deposit was formed on the anode using negatively charged powder suspensions. In this case the presence of bubbles was prevented by using a Cu electrode, which consumes the oxygen formed during the electrolysis. The alkaline suspension with the addition of TMAH resulted in a firm and dense deposit, containing 67 wt.% solids. A similar result was obtained if the pH was adjusted to 9 by adding NaOH. In contrast, no deposit was formed from the suspension with Dolapix until the pH was adjusted to 9, where the Dolapix is well dissociated and helps to produce dense and firm deposits. The above results suggest that for electrophoretic deposition the most appropriate suspensions for cathodic deposition were those containing CTAB, while for deposition on an anode, the addition of Dolapix and an increase in the pH seem to give the best results, since they lead to a high particle-packing density in the deposits. 3.3. Infiltration of SiC-fibre woven fabrics In the next set of experiments we explored the electrophoretic infiltration of SiC-fibre woven fabrics by SiC particles. Based on the above results we selected the two compositions with the highest ZP: the suspension of positively charged particles with a 0.5% CTAB addition (48 mV) and the suspension with negatively charged particles and Dolapix addition (−49 mV). For better understanding of the infiltration, the ZP of the SiC fibres was also analysed, as shown in Fig. 4. The as-received fibres used Fig. 4. Zeta-potential of non-treated (a) and (b) SDOSS-treated SiC fibres as a function of pH (the fibres were cut and crushed for the analysis)
S Novak et al. Journal of the European Ceramic Society 28 (2008)2801-2807 2805 in this study were seen to be negatively charged(Fig. 4, open was used as the anode. The infiltrated green samples were dried square symbols)exhibiting similar behaviour to the Sic pow- slowly and examined with an optical microscope. Further they der(see Fig. 1). The fibres were then treated with the anionic were vacuum infiltrated with a resin and examined by SEM to surfactant SDOSS in order to improve the wetting with aqueous estimate the success of the infiltration slurries, which also resulted in a slight increase in the negative Fig. 5a illustrates the as-received Sic-fibre fabric before ZP value(Fig. 4, close square symbols) infiltration, while Fig. 5b and c show the surfaces of the fab- The Sic-fibre fabric was first used as a deposition electrode, rics electrophoretically infiltrated with negatively or positively but no deposit was formed within 5 min. The current during the charged SiC particles, respectively. In the first case the fibres deposition was very low, probably due to the conductivity of the were pre-treated with SDOSs. In this case most of the large fibres being too low. Then, one or several layers of the fabric voids between the fibre tows were filled and the particles homo- were placed in front of the steel electrode, so that the particles geneously coated the surface of the fabric(Fig. 5b), while the were forced to travel through the fabric towards the electrode. positively charged suspension seems to flocculate on contact In order to prevent the formation of bubbles in the deposit a with the negatively charged fibres. sEM observations(Fig. 6a membrane was placed in front of the fabric at the cathode or, and b) suggest that in the case of positively charged particle in the case of negatively charged particles, a copper electrode they attach themselves to the fibres rather than entering the fibre tows interspaces. This is in agreement with a previously pro- by EPD, which indicates that a more efficient infiltration can be achieved by using equal-sign charges of the fibres and particles rather than using opposite charges A low-angle cross-section of a green part prepared with negatively charged suspension, infiltrated with a polymer and carefully polished, was also observed by' distinguish the means of sem as shown in Fig. 7, taking special care to Sic-infiltrated parts(the light phase in Fig. 7a) from the polymer- infiltrated parts(dark areas in Fig. 7a, see the arrows).As illustrated in Fig. 7b and c, which show the light areas, SiC particles have effectively filled the narrow gaps between the (a)圆 0,2mm Fig. 5. Optical micrographs of the starting fibre fabric (a), the fabric infiltrated with a TMAH-dispersed SiC suspension(b) and the fabric infiltrated with the Fig. 6. SEM micrographs of the surface of the fabric infiltrated with an alkaline CTAB-dispersed suspension(c) SiC suspension(a and b: different magnifications
S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 2805 in this study were seen to be negatively charged (Fig. 4, open square symbols) exhibiting similar behaviour to the SiC powder (see Fig. 1). The fibres were then treated with the anionic surfactant SDOSS in order to improve the wetting with aqueous slurries, which also resulted in a slight increase in the negative ZP value (Fig. 4, close square symbols). The SiC-fibre fabric was first used as a deposition electrode, but no deposit was formed within 5 min. The current during the deposition was very low, probably due to the conductivity of the fibres being too low. Then, one or several layers of the fabric were placed in front of the steel electrode, so that the particles were forced to travel through the fabric towards the electrode. In order to prevent the formation of bubbles in the deposit a membrane was placed in front of the fabric at the cathode or, in the case of negatively charged particles, a copper electrode Fig. 5. Optical micrographs of the starting fibre fabric (a), the fabric infiltrated with a TMAH-dispersed SiC suspension (b) and the fabric infiltrated with the CTAB-dispersed suspension (c). was used as the anode. The infiltrated green samples were dried slowly and examined with an optical microscope. Further they were vacuum infiltrated with a resin and examined by SEM to estimate the success of the infiltration. Fig. 5a illustrates the as-received SiC-fibre fabric before infiltration, while Fig. 5b and c show the surfaces of the fabrics electrophoretically infiltrated with negatively or positively charged SiC particles, respectively. In the first case the fibres were pre-treated with SDOSS. In this case most of the large voids between the fibre tows were filled and the particles homogeneously coated the surface of the fabric (Fig. 5b), while the positively charged suspension seems to flocculate on contact with the negatively charged fibres. SEM observations (Fig. 6a and b) suggest that in the case of positively charged particles, they attach themselves to the fibres rather than entering the fibre tows interspaces. This is in agreement with a previously proposed mechanism for particle infiltration of fibrous substrates by EPD, which indicates that a more efficient infiltration can be achieved by using equal-sign charges of the fibres and particles rather than using opposite charges.24 A low-angle cross-section of a green part prepared with a negatively charged suspension, infiltrated with a polymer and carefully polished, was also observed by means of SEM, as shown in Fig. 7, taking special care to distinguish the SiC-infiltrated parts (the light phase in Fig. 7a) from the polymerinfiltrated parts (dark areas in Fig. 7a, see the arrows). As illustrated in Fig. 7b and c, which show the light areas, SiC particles have effectively filled the narrow gaps between the Fig. 6. SEM micrographs of the surface of the fabric infiltrated with an alkaline SiC suspension (a and b: different magnifications)
2806 S. Novak et aL. /Journal of the European Ceramic Society 28 (2008)2801-2807 (a)ll For the powder used in this investigation the best results,i.e,a high zeta-potential leading to firm, dense deposits, were obtained for the addition of o5 wt %o of cationic deflocculant CtAB or by adjusting the pH of the suspension to 9 and adding Dolapix. The formation of bubbles due to electrolysis was prevented by using a membrane, in the case of cathodic deposition, and a Cu electrode, in the case of the anodic deposition. The infiltration of an SDOSS-pre-treated Tyrano SA Sic fibre fabric with a negatively charged SiC suspension resulted in dense fibre-based green parts, while the infiltration with suspension of positively charged particles was less effective Powder infiltration prior to the PlP process may considerably by avoiding the several required steps of polymer immersion. Electrophoretic infiltration may also significantly improve the density of SiC/SiC composites, in particular by filling the large voids between the fibre tows This should enable the fabrication of SiC/SiC composites with a close-to-zero porosity, with any remaining pores being closed pores, thus ensuring a hermetic Acknowledgements S Novak would like to thank the royal Society, UK, and the uropean Commission for their financial support of her stay at Imperial College, London, where a part of this work was per- formed. Mr J. Cho(Imperial College London)is acknowledged for experimental assistance The EC (EURATOM-Fusion) and the Agency of Science of the Republic of Slovenia are acknowledged for their financial support of the research on fusion-relevant composites. Dr. A. Ortona, NT Spa, Italy, and Dr B Riccardi (EFDA)are acknowl edged for the supply of the SiC-fibre fabric Special thanks go to Ms. Petra Ursnik for her sustained efforts in the laboratory References U- X27000 10Nm wDk Fig. 7. SEM micrographs of a polished low-angle cross-section of the fabric 1. Muroga, T, Gasparotto, M. and Zinkle, S.J., Overview of materials research for fusion reactors. Fusion Eng. Des., 2002, 13 infiltrated with an alkaline SiC suspension(a-c: different magnifications) 2. Tavassoli, A. A F, Present limits and improvements of structural materials for fusion reactors-a review. J. Nuclear Mater. 2002. 302. 73. 3. Naslain, R, Design, preparation and properties of non-oxide CMCs for individual fibres, demonstrating the success of the develope application in engines and nuclear reactors: an overview. Comp. Sci. Tech- no.,2004,64,155. 4. Riccardi, B. Giancarli, L, Hasegawa, A, Katoh, Y, Kohyama, A, Jones, R. H et al, Issues and advances in SiCf/SiC composites development for 4. Conclusions nead. L.L. and we This study has investigated the influence of surface charges on posites for fusion energy applications. J. Nuclear Mater, 2002, 307-311 the electrophoretic deposition of Sic powder in aqueous suspen 6. Scholz, H. W, Frias Rebelo, A.J. Rickerby, D. G, Krogul, P, Lee, w sions and has demonstrated the usefulness of the electrophoretic E, Evans, J. H. et al, Swelling behaviour and TEM studies of SiCf/SiC infiltration of Sic-fibre fabrics for production of SiC/Sic com omposites after fusion relevant helium implantation. J. Nuclear Mater 1998,258-263,1572. Based on the characteristics of the suspensions and of the bulk 7. Katoh, Y, Kohyama, A Nozawa, T and Sato, composites deposits formed on metallic electrodes, we confirmed that a high through transient eutectic-phase route for fusion 丿. Nuclear while the suspension condi importance for the EPD process, 8. Muroga, T, Gasparotto, M and Zinkle, S.J. Overview of materials research zeta-potential is of outmost ivity plays only a secondary role for fusion reactors. Fusion Eng Des., 2002, 61-62, 13
2806 S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 Fig. 7. SEM micrographs of a polished low-angle cross-section of the fabric infiltrated with an alkaline SiC suspension (a–c: different magnifications). individual fibres, demonstrating the success of the developed EPD technique. 4. Conclusions This study has investigated the influence of surface charges on the electrophoretic deposition of SiC powder in aqueous suspensions and has demonstrated the usefulness of the electrophoretic infiltration of SiC-fibre fabrics for production of SiC/SiC composites. Based on the characteristics of the suspensions and of the bulk deposits formed on metallic electrodes, we confirmed that a high zeta-potential is of outmost importance for the EPD process, while the suspension conductivity plays only a secondary role. For the powder used in this investigation the best results, i.e., a high zeta-potential leading to firm, dense deposits, were obtained for the addition of 0.5 wt.% of cationic deflocculant CTAB or by adjusting the pH of the suspension to 9 and adding Dolapix. The formation of bubbles due to electrolysis was prevented by using a membrane, in the case of cathodic deposition, and a Cu electrode, in the case of the anodic deposition. The infiltration of an SDOSS-pre-treated Tyrano SA SiC- fibre fabric with a negatively charged SiC suspension resulted in dense fibre-based green parts, while the infiltration with a suspension of positively charged particles was less effective. Powder infiltration prior to the PIP process may considerably shorten the densification time of SiC/SiC composite production by avoiding the several required steps of polymer immersion. Electrophoretic infiltration may also significantly improve the density of SiC/SiC composites, in particular by filling the large voids between the fibre tows. This should enable the fabrication of SiC/SiC composites with a close-to-zero porosity, with any remaining pores being closed pores, thus ensuring a hermetic material. Acknowledgements S. Novak would like to thank the Royal Society, UK, and the European Commission for their financial support of her stay at Imperial College, London, where a part of this work was performed. Mr. J. Cho (Imperial College London) is acknowledged for experimental assistance. The EC (EURATOM-Fusion) and the Agency of Science of the Republic of Slovenia are acknowledged for their financial support of the research on fusion-relevant composites. Dr. A. Ortona, NT Spa, Italy, and Dr. B. Riccardi (EFDA) are acknowledged for the supply of the SiC-fibre fabric. Special thanks go to Ms. Petra Ursnik for her sustained efforts in the laboratory ˇ work. References 1. Muroga, T., Gasparotto, M. and Zinkle, S. J., Overview of materials research for fusion reactors. Fusion Eng. Des., 2002, 13. 2. Tavassoli, A. A. F., Present limits and improvements of structural materials for fusion reactors—a review. J. Nuclear Mater., 2002, 302, 73. 3. Naslain, R., Design, preparation and properties of non-oxide CMCs for application in engines and nuclear reactors: an overview. Comp. Sci. Technol., 2004, 64, 155. 4. Riccardi, B., Giancarli, L., Hasegawa, A., Katoh, Y., Kohyama, A., Jones, R. H. et al., Issues and advances in SiCf/SiC composites development for fusion reactors. J. Nuclear Mater., 2004, 329–333, 56. 5. Snead, L. L. and Weber, W. J., Promise and challenges of SiCf/SiC composites for fusion energy applications. J. Nuclear Mater., 2002, 307–311, 1057. 6. Scholz, H. W., Frias Rebelo, A. J., Rickerby, D. G., Krogul, P., Lee, W. E., Evans, J. H. et al., Swelling behaviour and TEM studies of SiCf/SiC composites after fusion relevant helium implantation. J. Nuclear Mater., 1998, 258–263, 1572. 7. Katoh, Y., Kohyama, A., Nozawa, T. and Sato, M., SiC/SiC composites through transient eutectic-phase route for fusion applications. J. Nuclear Mater., 2004, 329–333, 587. 8. Muroga, T., Gasparotto, M. and Zinkle, S. J., Overview of materials research for fusion reactors. Fusion Eng. Des., 2002, 61–62, 13
S Novak et al. Journal of the European Ceramic Society 28 (2008)2801-2807 2807 9. Nannetti, C A, Ortona, A, de Pinto, D A and Riccardi, B, Manufacturing 17. Novak, S, Mejak, K and Drazic, G, The preparation of LPS SiC-fibre- Sic-fiber-reinforced Sic matrix composites by improved CVI/slurry infil- reinforced SiC ceramics using electrophoretic deposition. J. Mater. Sc ration/polymer impregnation and pyrolysis. J. Am. Ceram. Soc., 2004, 87, 2006,41,8093 205 18. Novak, S, Drazic, G. and Mejak, K, Electrophoretic deposition of green 10. Heavens, S.N., Electrophoretic deposition as a processing route for ceram parts for LPS SiC-based ceramics. Key Eng Mate, 2006, 314, 45 ics. In Advanced Ceramic Processing and Technology, ed J. G. P. Binner. 19. S. Novak, K. Konig, Electrophoretic deposition of alumina parts from Noyes Publications, Pask Ridge, New Jersey, 1990, P. 255. ethanol and 11. Van der Biest, O and Vandeperre, L J, Electrophoretic deposition of mate- 20. Toplisek, T, Drazic, G, Novak, S and Kobe, S, Electron microscopy and rials. Anny. rey Mater Sci. 1999.29.327 microanalysis of the fiber/matrix interface in SiC-based ceramic composite 12. Sarkar, P and Nicholson, P. S, Electrophoretic deposition(EPD): mecha- material for use in a fusion reactor application. Scanning, 2008, 30, 35 nisms, kinetics, and application to ceramics. J. Am. Ceram. Soc., 1996, 79, 21. Zhang, J, Xu, Q, Feng Ye, Lin, Q, Jiang, D and Iwasa, M., Effect of citric 1987. acid on the adsorption behavior of polyethylene (PED and the relevant 13. Fukada, Y, Nagarajan, N, Mekky, w, Bao, Y. Kim, H.S. and Nicholson, P. stability of SiC slurries. Colloids Surf. A: Physicochem. Eng. Asp, 2006. S, Electrophoretic deposition-mechanisms, myths and materials. J Mater 276,168. Sci,2004,39,787. 22. Zhang, J and Iwasa, M, Dispersion of SiC in aqueous media with Al2O 14. Boccaccini, R and Zhitomirsky, I, Application of electrophoretic and elec- and Y203 as sintering additives. J. Am. Ceram. Soc., 2005, 88, 1013 rolytic deposition techniques in ceramics processing. Curr. Opin. Solid State 23. Popa, A.M., Vleugels, J, Vermant, J and Van der Biest, O, Influence of sur- Mater: Sci.. 2002. 6. 251 factant addition sequence on the suspension properties and electrophoretic 15. Boccaccini, R, Kaya, C and Chawla, K. K, Use of electrophoretic depo- deposition behaviour of alumina and zirconia. J. Eur. Ceram Soc, 2006, sition in the processing of fibre reinforced ceramic and glass m 6.933 46 composites. A review. Compos. Part A, 2001, 32, 997 24. Stoll, E, Mahr, P. Kruger. H -G.Kern. H, Thomas, B. J. C. and Bo Moritz, K and Muller, E, Electrophoretic infiltration of woven carbon fibre caccini,AR, Fabrication technologies for oxide-oxide ceramic matrix mats with SiC powder suspensions. Euro Ceramics VIl, PT 1-3. Key Eng ectrophoretic deposition J. Eur. Ceram. Soc., 2006. later,2002,206,19
S. Novak et al. / Journal of the European Ceramic Society 28 (2008) 2801–2807 2807 9. Nannetti, C. A., Ortona, A., de Pinto, D. A. and Riccardi, B., Manufacturing SiC-fiber-reinforced SiC matrix composites by improved CVI/slurry infiltration/polymer impregnation and pyrolysis. J. Am. Ceram. Soc., 2004, 87, 1205. 10. Heavens, S. N., Electrophoretic deposition as a processing route for ceramics. In Advanced Ceramic Processing and Technology, ed. J. G. P. Binner. Noyes Publications, Pask Ridge, New Jersey, 1990, p. 255. 11. Van der Biest, O. and Vandeperre, L. J., Electrophoretic deposition of materials. Annu. Rev. Mater. Sci., 1999, 29, 327. 12. Sarkar, P. and Nicholson, P. S., Electrophoretic deposition (EPD): mechanisms, kinetics, and application to ceramics. J. Am. Ceram. Soc., 1996, 79, 1987. 13. Fukada, Y., Nagarajan, N., Mekky, W., Bao, Y., Kim, H. S. and Nicholson, P. S., Electrophoretic deposition—mechanisms, myths and materials. J. Mater. Sci., 2004, 39, 787. 14. Boccaccini, R. and Zhitomirsky, I., Application of electrophoretic and electrolytic deposition techniques in ceramics processing.Curr. Opin. Solid State Mater. Sci., 2002, 6, 251. 15. Boccaccini, R., Kaya, C. and Chawla, K. K., Use of electrophoretic deposition in the processing of fibre reinforced ceramic and glass matrix composites. A review. Compos. Part A, 2001, 32, 997. 16. Moritz, K. and Muller, E., Electrophoretic infiltration of woven carbon fibre mats with SiC powder suspensions. Euro Ceramics VII, PT 1–3. Key Eng. Mater., 2002, 206, 193. 17. Novak, S., Mejak, K. and Draziˇ c, G., The preparation of LPS SiC-fibre- ´ reinforced SiC ceramics using electrophoretic deposition. J. Mater. Sci., 2006, 41, 8093. 18. Novak, S., Draziˇ c, G. and Mejak, K., Electrophoretic deposition of green ´ parts for LPS SiC-based ceramics. Key Eng. Mater., 2006, 314, 45. 19. S. Novak, K. Konig, Electrophoretic deposition of alumina parts from ¨ ethanol and aqueous suspensions, in preparation. 20. Toplisek, T., Dra ˇ ziˇ c, G., Novak, S. and Kobe, S., Electron microscopy and ´ microanalysis of the fiber/matrix interface in SiC-based ceramic composite material for use in a fusion reactor application. Scanning, 2008, 30, 35. 21. Zhang, J., Xu, Q., Feng Ye, Lin, Q., Jiang, D. and Iwasa, M., Effect of citric acid on the adsorption behavior of polyethylene imine (PEI) and the relevant stability of SiC slurries. Colloids Surf. A: Physicochem. Eng. Asp., 2006, 276, 168. 22. Zhang, J. and Iwasa, M., Dispersion of SiC in aqueous media with Al2O3 and Y2O3 as sintering additives. J. Am. Ceram. Soc., 2005, 88, 1013. 23. Popa, A. M., Vleugels, J., Vermant, J. and Van der Biest, O., Influence of surfactant addition sequence on the suspension properties and electrophoretic deposition behaviour of alumina and zirconia. J. Eur. Ceram. Soc., 2006, 26, 933. 24. Stoll, E., Mahr, P., Kruger, H.-G., Kern, H., Thomas, B. J. C. and Boc- ¨ caccini, A. R., Fabrication technologies for oxide–oxide ceramic matrix composites based on electrophoretic deposition. J. Eur. Ceram. Soc., 2006, 26, 1567